arXiv subjects

Reza Soosahabi

Publications and source records attributed to Reza Soosahabi.

3 recordsLinked to original sources

arXiv2026

Evaluation and Hardening of LLM System Instructions Against Extraction via Encoding Attacks

System Instructions in Large Language Models (LLMs) are commonly used to enforce safety policies, define agent behavior, and protect sensitive operational context in agentic AI applications. These instructions may contain sensitive information such as API credentials, internal policies, and privileged workflow definitions, making system instruction leakage a critical security risk highlighted in the OWASP Top 10 for LLM Applications. Without incurring the overhead costs of reasoning models, many LLM applications rely on refusal-based instructions that block direct requests for system instructions, implicitly assuming that prohibited information can only be extracted through explicit queries. We introduce an automated evaluation framework that tests whether system instructions remain confidential when extraction requests are re-framed as encoding or structured output tasks. Across seven common models and 46 verified system instructions, we observe high attack success rates (>0.7) for structured serialization where models refuse direct extraction requests but disclose protected content in the requested serialization formats. We further demonstrate a mitigation strategy based on one-shot instruction reshaping using a Chain-of-Thought reasoning model, indicating that even subtle changes in wording and structure of system instructions can significantly reduce attack success rate without requiring model retraining.

cs.CR↗

arXiv2026

Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems

Agentic AI systems increasingly rely on language-model components to interpret instructions, process external data, invoke tools, and coordinate with other agents. These capabilities make prompt-injection and jailbreak attacks more consequential, especially as attackers adopt model-guided automation to scale probing, prompt refinement, and response evaluation. This work analyzes the resulting attack-defense setting through a probabilistic model of a target system, its defense mechanism, and the attacker's automated judge. Our analysis shows that conventional detect-and-block defenses can allow attacker success rate (ASR) to approach one as the query budget grows, since predictable refusals provide useful feedback to automated search. We then examine detect-and-misdirect, where detected malicious interactions receive controlled, non-operational responses designed to induce false-positive errors in the attacker's judge. This strategy reduces the positive predictive value of attacker-selected candidates and yields a bounded asymptotic ASR. We evaluate a proof-of-concept realization of this strategy through Contextual Misdirection via Progressive Engagement (CMPE), a lightweight conversational misdirection method designed to replace predictable refusal text with safe but strategically misleading responses in automated jailbreak settings. On jailbreak benchmarks, CMPE reduces estimated ASR upper bounds by up to two orders of magnitude and nearly eliminates verified attack success in end-to-end PAIR and GPTFuzz attack runs.

cs.CR↗

arXiv2022

On Securing MAC Layer Broadcast Signals Against Covert Channel Exploitation in 5G, 6G & Beyond

In this work, we propose a novel framework to identify and mitigate a recently disclosed covert channel scheme exploiting unprotected broadcast messages in cellular MAC layer protocols. Examples of covert channel are used in data exfiltration, remote command-and-control (CnC) and espionage. Responsibly disclosed to GSMA (CVD-2021-0045), the SPARROW covert channel scheme exploits the downlink power of LTE/5G base-stations that broadcast contention resolution identity (CRI) from any anonymous device according to the 3GPP standards. Thus, the SPARROW devices can covertly relay short messages across long-distance which can be potentially harmful to critical infrastructure. The SPARROW schemes can also complement the solutions for long-range M2M applications. This work investigates the security vs. performance trade-off in CRI-based contention resolution mechanisms. Then it offers a rigorously designed method to randomly obfuscate CRI broadcast in future 5G/6G standards. Compared to CRI length reduction, the proposed method achieves considerable protection against SPARROW exploitation with less impact on the random-access performance as shown in the numerical results.

cs.CR↗